July 6, 2019

Preview

  1. Introduction to single-cell RNA-seq
  2. Quality control and normalization
  3. Survey of downstream analysis methodology

Transferring principles and concepts from bulk

  • Count distributions with overdispersion
  • Variance stabilizing transformations
  • Normalization for sequencing depth
  • Removal of batch effects

Key features of single-cell

  • Heterogeneity/increased resolution
  • Sparsity (few genes detected in each cell)
  • New technical artifacts
    • Different protocols
    • Smaller starting material
    • ++ Multiplexing

Increased resolution in single-cell RNA-seq

Bulk RNA-seq measures the average expression across a pool of cells

Sparsity in single-cell RNA-seq

  • Detection rate = proportion of genes with nonzero measurement for that cell
  • Often the vast majority of measurements are zero
  • Commonly referred to as “dropout” (1-detection rate)

New techinical artifacts

Evolution of single-cell technology

Terminology: platforms vs protocols

  • platforms isolate cells, extract mRNA, and prepare libraries for sequencing
  • protocols refer to the specific chemistries used to prepare sequencing libraries
  • platforms may be protocol-specific (proprietary) or allow multiple protocols (open-source)
Drop-seq    vs    10x    

Main platform/protocol distinctions

  • Full-length transcript vs 3’ end only
  • Plate-based vs droplet-based
  • Read counts vs UMIs
  • Multiplexed with barcodes vs not

Full-length transcript vs 3’ end only

Isoform analysis only possible with full-length

Plate-based

Droplet-based (higher throughput)

Unique Molecular Identifiers (UMIs)

  • PCR introduces nonlinear amplification bias
  • UMIs are a way to tag each unique molecule in the sequencing library (before PCR)
  • Number of possible UMIs = \(4^L\), where \(L\) is the length of the UMI
  • Low \(L\) or sequencing errors can cause barcode collisions
  • Afterward, count up only the number of distinct UMIs (collapse reads)

Multiplexing

Beads

Spike-ins

Tradeoff between capture rate and read depth